R Faceted qqplots with column and row

问题

I needs to create a qqplot using facets by row and column. I understand how to facet plot with columns and rows, but I am not sure how to set up my data. Ultimately, I want to group my dataset by column and row, then sort the 'Modeled' results and 'Observed' results in ascending order while adding a column with the 'row' group and a column with the 'column' group.

I have been trying to modify the solution to this question, Faceted qqplots with ggplot2, , but I am not very familiar with lapply so maybe I just missed something.

Here is the code I have been working with:

#Dummy Data:
df <- mtcars
# Name columns as I have in my real data
df$rows <- df$cyl
df$columns <- df$gear
df$Modeled <- df$wt
df$Observed <- df$mpg

# Function to sort data while maintaining the rows & columns for use in facet later.
dat_sort <- do.call("rbind", 
                  sapply(list(unique(df$rows), unique(df$columns)), 
                         FUN = function(x) {
                           data.frame(rows = x[[1]], 
                                      columns = x[[2]], 
                                      Observed = sort(df$Observed[df$rows == x[[1]] & df$columns == x[[2]]]), 
                                      Modeled = sort(df$Modeled[df$rows == x[[1]] & df$columns == x[[2]]])
                                      )
                           }
                         ))

I don't get an error, but my output is definitely not what I was expecting. My output should look like this: (with correct column names)

rows columns Observed Modeled
6      4      17.8      2.620
6      4      19.2      2.875
6      4      21.0      3.440
6      4      21.0      3.440
4      3      21.5      2.465
8      5      15.0      3.17
8      5      15.8      3.57

Output from code:

       [,1]   [,2]   [,3]   [,4]
[1,]  6.000  6.000  6.000  6.000
[2,]  4.000  4.000  4.000  4.000
[3,] 17.800 19.200 21.000 21.000
[4,]  2.620  2.875  3.440  3.440
[5,]  4.000  4.000  4.000  4.000
[6,]  3.000  3.000  3.000  3.000
[7,] 21.500 21.500 21.500 21.500
[8,]  2.465  2.465  2.465  2.465

Any help would be most appreciated!

Thanks!

回答1:

We just need to order the dataset in ascending order and that can be done with order

df1 <- df[c('rows', 'columns', 'Observed', 'Modeled')]
df2 <- df1[do.call(order, -df1),]

In the OP's code, change the sapply to lapply to return as list (sapply by default uses simplify = TRUE to return a matrix)

do.call(rbind, lapply(list(unique(df$rows),
        unique(df$columns)), 
            FUN = function(x) {
               data.frame(rows = x[[1]], 
                          columns = x[[2]], 
                          Observed = sort(df$Observed[df$rows == x[[1]] & 
                                   df$columns == x[[2]]]), 

                          Modeled = sort(df$Modeled[df$rows == x[[1]] &
                                     df$columns == x[[2]]])
                                                               )
                                                    }
                                                  ))
#   rows columns Observed Modeled
#1    6       4     17.8   2.620
#2    6       4     19.2   2.875
#3    6       4     21.0   3.440
#4    6       4     21.0   3.440
#5    4       3     21.5   2.465

By doing the looping on the unique as a list, each vector is a separate list element and x[[1]], x[[2]] are actually subsetting based on the 6, 4 and 4, 3 for the second element

list(unique(df$rows), unique(df$columns))
#[[1]]
#[1] 6 4 8

#[[2]]
#[1] 4 3 5

Instead, if we need to do this for corresponding elements, then use Map or loop over the sequence of unique elements (assuming they have the same length), but a more easier approach is split

If the lengths are same for unique elements, and want to subset based on corresponding values, then use Map

do.call(rbind, Map(function(x, y) {
    i1 <- df$rows == x & df$columns == y
    data.frame(rows = x, columns = y,
               Observed = sort(df$Observed[i1]),
               Modeled = sort(df$Modeled[i1]))},
    unique(df$rows), unique(df$columns)))
#  rows columns Observed Modeled
#1    6       4     17.8   2.620
#2    6       4     19.2   2.875
#3    6       4     21.0   3.440
#4    6       4     21.0   3.440
#5    4       3     21.5   2.465
#6    8       5     15.0   3.170
#7    8       5     15.8   3.570

回答2:

transpose to matrix:

t(dat_sort)
     [,1] [,2] [,3]  [,4] [,5] [,6] [,7]  [,8]
[1,]    6    4 17.8 2.620    4    3 21.5 2.465
[2,]    6    4 19.2 2.875    4    3 21.5 2.465
[3,]    6    4 21.0 3.440    4    3 21.5 2.465
[4,]    6    4 21.0 3.440    4    3 21.5 2.465

来源：https://stackoverflow.com/questions/61963801/r-faceted-qqplots-with-column-and-row

标签